ccHarmony

ccHarmony数据集是由上海交通大学创建的，专注于图像和谐化处理，旨在通过调整合成图像中的前景使其与背景更加协调。该数据集基于现有包含光照信息的数据集，通过转换真实图像中的前景光照条件来创建合成图像。数据集包含4260对合成图像和真实图像，主要用于训练深度学习模型以提高图像和谐化性能。创建过程中，利用了NUS和Gehler数据集中的光照信息，通过多项式匹配矩阵进行颜色转换。ccHarmony数据集的应用领域主要集中在图像编辑和增强，特别是在解决合成图像中光照不一致的问题上。

The ccHarmony dataset was created by Shanghai Jiao Tong University, focusing on image harmonization. Its core goal is to adjust the foreground in composite images to make it more consistent with the background. This dataset is developed based on existing datasets containing illumination information, and composite images are generated by converting the foreground illumination conditions from real images. It includes 4260 pairs of composite images and real images, and is primarily used for training deep learning models to improve the performance of image harmonization. During the dataset construction process, illumination information from the NUS and Gehler datasets was utilized, with color conversion performed via a polynomial matching matrix. The application scenarios of the ccHarmony dataset mainly concentrate on image editing and enhancement, particularly in addressing the issue of illumination inconsistency in composite images.